Payments platform Nuggets has teamed up with the @bankofengland to work on a privacy and identity layer for a potential digital pound.
@camomileshumbs reports
https://t.co/I6ElPAMUxk
Access decides what systems your agent can reach. Authority decides whether this specific action, for this user, right now, is one it's allowed to take.
We just shipped langchain-nuggets, for LangChain and LangGraph developers. It enforces that at execution time and leaves a proof anyone can verify, even across orgs and clouds.
Trust travels with the agent. Every allowed action carries a receipt a partner or a regulator on a different cloud can verify for themselves, without access to your systems, without trusting your logs, and without exposing your raw data. Hashes, not values.
See it running: the cross-org demo, locally, with one ALLOW, four DENY paths, intent binding, and proof verification. https://t.co/VZonCuUqCJ
Open source. Python and TypeScript.
pip install langchain-nuggets
npm install @nuggetslife/langchain-nuggets
Code: https://t.co/lGO3iIOZtk
Full write-up: https://t.co/yFEyWyNvmx
Proud to be named in @techrounduk's AI45 2026, celebrating the most exciting AI companies across the world!
A big thanks to the judges and congrats to everyone on the list. Full list 👇
https://t.co/lAigdddwmS
Who is acting? On whose authority? Under what constraints, and how do we prove it?
These are the questions that matter when you deploy AI Agents.
To help Enterprises make the right procurement decision, we’ve created a checklist. Free to use.
https://t.co/zuOsqTdYK3
2/2 We cover the five questions it closes with, and the one that comes before any of them: should the agent be allowed to act at all, on whose authority, under what constraints, and where is the proof.
https://t.co/3O4D9nhd38
1/2 Rubrik ZeroLabs surveyed IT managers this week. 77% say their AI agents are out of control. 86% expect proliferation to outpace their guardrails inside a year.
The piece reads like a description of the problem Nuggets solves.
On stage Wed at @WomeninPayments UK-EU 2026.
Empowering Women in Tech, with Sarah Bennett (@BCG ), Nathalie Oestmann (Mangosteen) and Nina Sametinger (@PayPal ).
Sharing what I've learned, including what I'd tell a younger me.
Procurement is being asked to sign off on AI systems, that their frameworks were never designed for.
18 questions to ask. Free to use.
https://t.co/zuOsqTdYK3
Most teams evaluating AI systems don't share a common language for what those systems actually do.
We built one.
Plain-language glossary, free to use.
https://t.co/bulA6IP7NE
Read our latest blog on the Enterprise AI Governance Framework.
If you're deploying autonomous AI: how do you know the AI action was authorised?
Most enterprises can't. Here's where to start.
https://t.co/wajock9vaA
Most companies can control who gets into their AI systems.
Almost none can prove what those systems did once they were in.
@BiometricUpdate covered our Enterprise AI Governance Framework & why the control point for AI has shifted from access to execution.
https://t.co/WFTctI1vDv
We've published the Enterprise AI Governance Framework.
Most organisations can control who gets in. Nobody can prove what AI does once it's there.
The control point has shifted from access to execution.
AI Systems Need Governance, Not Just Security
A standard vulnerability. An autonomous agent. Two hours to access millions of internal records.
Security wasn't the gap. Governance was.
https://t.co/IlQfqPtkse #AIGovernance
You Don't Have to Choose Between Privacy and Compliance
The breach cycle continues. Same response every time - more tooling, stronger defences, repeat.
The problem isn't the security. It's centralised data storage - and there's an alternative.
https://t.co/U8kNcgU1B4
Nuggets is now part of #NVIDIAConnect.
As AI agents move from pilots to production, compute isn’t enough.
Every action needs verifiable identity, delegated authority, and policy bound at execution.
That’s the trust infrastructure production AI requires.
AI agents don't log in. They execute - across clouds, tools, and orgs, without supervision.
Pilots tolerate ambiguity. Production doesn't.
Authority must be bound to every action at execution. Not inferred later.
↓ https://t.co/B5dytRLpC9
@steipete Great. We built a privacy-preserving KYA/KYC plugin for @elizaOS, and we’re exploring how to reuse that work to support Clawdbot deployments. Focus is on identity, ownership, and accountability, not redesigning agents 🙏🏽
https://t.co/uX3hDAIXIT